Research Of Retinal Image Segmentation Based On Fuzzy Clustering And Morphology Filters

The retinal vessel detection and network structure analysis based on digital image have important application value on the field of information security and medical diagnosis. The process of retinal vessel testing is the process of image segmentation indeed. Through the retinal image segmentation, we can obtain the complete vascular network which can be considered the foundation of further analysis the network topology and geometric structure. Image segmentation is just to classify the pixels with similar principle. In morphology theory, in order to extract the corresponding shape and outline, we can use the different "structure elements" to measure the target area in image. In this paper, the segmentation method based on fuzzy clustering combine with morphology theory is used to extract the vessel in retinal image. The paper focuses on the research of retinal image segmentation based on fuzzy clustering and morphology filters. Major contribution is as follows:1. Discussed and analyzed the basic theory of fuzzy c-means clustering algorithm and it's properties, meanwhile studied the lots of problems while using FCM for image segmentation,such an getting the number of clusters, setting initial cluster centers and membership functions.2. Explains the mathematical morphology in detail, and analysis the basic theory of binary morphology which is including dilation, erosion, open operation and close operation. It also researched the application of morphology in the image processing, such as edge extraction, noise filter and regional fill etc.3. According to the characteristic in the retinal image such as: illumination uneven and the value of contrast are lower, an extraction method of retinal vessels based on fuzzy clustering in combination with morphological filtering is proposed. It is also using the algorithm to segment the retinal image which collected in DRIVE database for experiment, and verified the effectiveness of algorithm.